Tag Archives: Exposure

Shoot RAW, Reduce Noise?

Great Blue Heron

Great Blue Heron

Shoot RAW and you’ll have less noise.  OK, you and I both know that among RAW’s many benefits over JPEG, lower noise isn’t one of them.  Or is it?  Well, yes and no.

Technically, the level of noise you see in a JPEG can (should) be exactly the same as RAW, depending on the RAW conversion, not on JPEG’s inherent noise.  But look at the following pictures, noticing specifically the decreasing noise levels between them:  (These are screen shots pulled out of Aperture, thus the red highlighting in a few places.)

Two stop underexposed, adjusted in post

Two stop underexposed, adjusted in post

One stop underexposed, adjusted in post

One stop underexposed, adjusted in post

Correct exposure

Indicated exposure, adjusted in post

(Technical note:  All three shots are identical in camera except for exposure.  All three RAW files were adjusted identically except for exposure and recovery.  D300 @ ISO 400.  1/2500@f/5.6, 1/1250@f/5.6, 1/640@f/5.6)

Due to the Heron’s bright white feathers, the first shot is the greatest exposure I could get by with without blowing them out.  It’s actually a 2 stop underexposure compared to what the meter suggested.  Then in Aperture I gave it one stop boost and brought back the highlights with recovery.  In effect, this is the best I could have done with a JPEG.  (In fact the shadows are better due to the greater dynamic range maintained in the RAW file.  JPEG would not have had the same amount of data available.)  Note the very noticeable noise in the background.  Also, note in the pupils (and in many of the other background areas not visible in the crop) you can see a lot of dark areas that are pure black.

The second shot was given an additional stop in the camera.  This is the correct exposure for most of the scene, with the white feathers being the major exception.  So, in Aperture I left the exposure flat and just dialed in the same amount of recovery as the first one to regain detail in the white feathers.  That couldn’t have been done with a JPEG, as there would have been no data available to help out the feathers.  Note the much lower noise levels.  Also, all of the dark areas have detail in them.

Finally, the third shot was exposed as the camera meter suggested, giving a one stop overexposure to the scene.  In Aperture I pulled back the exposure by one full stop and matched the same recovery setting of the other two pictures.  As expected this gives a near exact exposure match for the other two.  The 14-bit RAW file from the D300 sensor has plenty of headroom to handle the highlights which are blown out by 1.75 stops, so I am able to bring them back to be basically the same as in the other two exposures.  The real benefit here, though, is the further reduction in noise.  It looks pretty good!

So does RAW have less noise than JPEG?  No.  But it does give you the headroom to overexpose by one, one and a half, maybe even 2 stops and then pull the exposure back in post, effectively reducing the noise levels and increasing the amount and quality of the detail in your shadow areas.

Had I thought of it at the time I would have also shot this at ISO 200 to determine if the overexposed shot (final shot above) had lower noise than I would have gotten simply by reducing the ISO.  The difference seems more dramatic than the nearly invisible difference between ISO 200 and 400, so I think this is better, but I’ll have to try it out to be sure.

Posted in Blog Posts Also tagged , , , , , , , |

Notes on Shooting HDR


(This post is actually forum conversation with a coworker. I thought the content might be useful for others.)

A few notes I’ve learned as I’ve tried to figure HDR out. I’ve started turning EVERYTHING on manual:  focus, white balance, and exposure. I used to stay in my usual aperture-priority mode, but noticed on my D200 that when I do auto-bracketing, it seems to do a new exposure reading between each exposure and I could get shots that were not really -2, -1, 0, +1, +2, but maybe -2, -2/3, 0, +1 1/3, +2. Photomatix seems to take it in stride, but it still bugs me so I fix it by setting the exposure to manual and then doing the autobracket.

And yes, I agree that having plus and minus two stops usually works out best. I would like to have a two-stop increment option on the autobracketing so I could just do a -2, 0, +2. I’ve seen a comparison between images using those three images and the same HDR tonemap that includes the -1 and +1 and there’s very little difference.

Something else I’ve found is that I need a camera with a faster frame rate. Like in the shot above.

It was hand-held, which I don’t like to do, but not only was there camera movement between shots that Photomatix handled fairly well, but there was also movement in the river and boat, which is really bothersome. I guess I could take the tonemapped shot into Photoshop with one or two of the individual shots and mask in those areas like some of the best HDR shooters do, but wow, what a pain! I think the best solution would just be to buy a D3 and crank the shots in twice as fast.

Posted in Conversation Also tagged , , |

Q&A: Fast Primes vs. Zooms

(This post is from an email conversation with two friends. I thought the content might be useful for others.)

Q:  [Friend 1] and I have talked in the past about brighter lenses. And at the time, he and I both had a theory that a brighter lens took a better (brighter) picture even at the same settings as a not-as-bright lens. So as an example, a 50mm prime @ f5.6 took a better picture than an 18-200 set at 50mm (or is that 75mm?) @ f5.6. What are your thoughts? I haven’t spent any time testing this theory in a methodical way, but I still have a gut feeling that it’s true.

A:  Lenses have two different metrics for defining how much light they are letting through. The one we’re familiar with is the f-stop. It is simply the focal length of the lens divided by the effective diameter of the lens. That’s why it’s written f/2.0 for instance. The lens’ aperture is f (the focal length) divided by two. What that means is that any simple lens whose ratio of the focal length to the lens stop is the same as another’s, regardless of absolute focal length or lens diameter, will let through the same exact amount of light.

Then there’s the T-stop. T stands for transmission. The above paragraph was about theoretical light transmission. The T-stop is the real transmission of a lens. The T-stop deviates from the f-stop mainly because of the number of elements within a lens. Whenever light travels through a surface of a lens, a small amount of the light is reflected away. With normal uncoated glass surfaces and refractive indices, that amount can be up to 5-10% of the light per surface. That causes two different issues. One is that your image is measurably dimmer than it should be, especially when you have a number of elements. Second is that reflected light has to go somewhere, and where it goes is both random (reduced contrast) and focused (flare) depending on the other elements and internal blackening of the lens.

The fix for that is multicoating on the lens surfaces (which can reduce the reflectance to less than 1% per surface) and minimizing the number of elements.

Your theory is correct that some lenses are brighter than others at the same f-stop. What causes it, though, is not the different speed of the lenses, but the number of elements in them, type and number of layers of coating, and lens design.

Thus, if you compare two hypothetical lenses, a 50mm f/1.4 and a 50mm f/2.8 (focal length doesn’t actually matter here, but if you were comparing pictures of the same scene side-by-side, it would be best to use identical focal lengths to make the comparison easier), that both have six multi-coated elements, pictures taken with both of them at f/5.6 should be identically exposed.

But if you took your 50/1.4 and your 18-200 @ 50mm and took an identically exposed photograph at f/5.6, it is very likely that the one with the zoom would be noticeably darker, possibly by half a stop or more. That is because the 50mm prime has seven multicoated elements and the 18-200 has 16 multicoated elements.

I haven’t read a lens review in a magazine in a long time, but I remember that they used to list the transmission of the lens so you could see the actual light loss.

Note that all of this stuff is automatically compensated for whenever you use your TTL (through the lens) meter. Although the exposure may be different between the two lenses at the same f-stop due to this T-stop reality, your TTL meter is seeing the scene through the two different lenses, so it will expose either of them correctly.

Lenses that are designed for use with studio lighting, and / or hand-held meters (like movie camera lenses) are even marked in T-stops instead of f-stops so that the meter reading from a hand-held meter will result in a correct and consistent exposure across different lenses.

Posted in Conversation Also tagged , , , , , |

Q&A: When Black Isn’t Black

(This post is from an email conversation with two friends. I thought the content might be useful for others.)

Q:  Anybody know what’s causing the black outfits to show their true colors?  To the eye, of course, these were simply black outfits.  All four images are taken with my Nikon D200, although I saw similar variations on my D70.  The top two were intentionally underexposed — as measured by the camera’s meter (which always want to overexpose in these kinds of situations, which would blow out the faces).  And the top two were both shot with a 50mm f/1.4 lens whereas the bottom two were shot with a 70-200mm f/2.8 lens.  Could it be coating on the lens? Photoshop CS3 interpreted them the same way Aperture did, so I really think it’s in the image data.

A:  Well, I would suggest that in our normal experience, there is no such thing as black, only shades of very dark shades of gray.  (I’ll not contemplate the philosophical implications with which I’m sure I disagree…)  And therefore anything dark we normally come across, with enough exposure, will become lighter and even white.  And whenever it gets enough exposure not to appear black, then it will possibly become non-neutral.  My wife is always asking me if her black shirt and pants “match”, and sometimes they don’t, but it really depends upon the light as to whether or not it’s evident.  This issue is similar to what causes metamerism (http://en.wikipedia.org/wiki/Metamerism_(color)) in B/W prints printed on most inkjet printers.

I can’t tell if the lighting is typical colored (RGB or RGBY) stage lighting.  If it was, then that will emphasize the colors of “neutral” fabrics even more.   All of this, plus the fact that different lighting has varying amounts of UV and IR light that we don’t necessarily see directly but that sometimes will be visible when reflected from stuff and that definitely will be picked up by digital camera sensors.

I doubt that the cause is due to the use of different Nikon lenses.  Nikon says about their SIC (Super Integrated Coating) that it is intended to maintain a “uniform color balance that characterizes Nikkor lenses.”  I’ve never noticed any particular Nikon brand lens having a color cast different from the rest.  (http://www.nikonimaging.com/global/products/lens/glossary.htm)

Most folk wouldn’t know to intentionally underexpose a dark subject and likely would have seen the issue even more than you did.

Posted in Conversation Also tagged , |